What voting system(s) do you like? Poll for single-seat elections

[lpetrich]

I wish to ask what voting systems you people like. Here, I will ask what you might like for single-seat positions, like president or governor or mayor or US/state senator or representative. You can vote for more than one system here. Here are the systems:

First past the post Vote for one candidate, and whoever gets the most votes wins.

Approval voting Vote for whichever candidates you like; you can vote for as many candidates as you want. Whoever gets the most votes wins.

Top-two runoff Vote in two rounds. The first round: for any one of all of the candidates. The second round: for one of the top two of the first round. Whoever gets the most second-round votes wins.

Preference voting Rank the candidates by your preference: which one is your first choice, your second choice, etc. With all the ballots, one uses some algorithm to find the overall winner. A common one is instant runoff: if nobody gets a majority of top votes, then the candidate with the fewest top votes is removed and the count is redone with that candidate's absence. Continue removing losing candidates until one candidate gets a majority. There are other algorithms that one can use, like Borda, Schulze, Tideman, etc.

Magical brownies In case you are all confused.

 

[DBT]

I lean towards Preference voting. Probably because it allows me to classify (perceived value) the candidates and place them into that order.

 

[bilby]

I lean towards Preference voting. Probably because it allows me to classify (perceived value) the candidates and place them into that order.

Me too.

It allows you to vote against candidates you dislike, as well as for candidates you like (if any). I always number my paper from the candidate I hate most through to the least objectionable.

(It is truly frightening in the Senate ballot to find that the Fred Nile Group are often not even in the bottom ten...)

 

[bigfield]

Preference voting, although few Australians seem to know how it works despite it being used for our federal and state systems.

 

[dx713]

Having seen first hand how two turns does (not) work (my country system), and feeling that approval would raise the same problems (do I support most my preferred candidate and risk a candidate I don't want passing, or support a consensus candidate and undermine my preferred candidate voice/visibility?), I think preference is the way to go.

IRV seems a good algorithm to me, close enough to giving the Condorcet candidate and not too complicated that it can't be explained step-by-step to someone with some basic education.
Alas, our politicians, even those who would benefit like the greens or other smaller parties, seem so oblivious to the existence of other systems than two-turns that they don't even use them within their own primaries... I'm not holding my breath. (and I hate the UK for having voted against such a system, that would have made a good foothold/demon for it in Western Europe)

 

[Nice Squirrel]

Preferential. There are times when I know at least two of the candidates and would like to support them both (aka both would be excellent at the job). Hell, I've even voted for Republican candidates on preferential ballots.

 

[Blahface]

The thing I don't like about rank choice voting(especially IRV) is that too many people just don't seem to understand it. Even if you can come up with a perfect Condorcet method in which there is no strategic benefit to burying your favorite candidate's top rival, you have to get the average voter to understand that. It might be tough to get the average voter to even understand what a Condorcet method is in the first place. Also, IRV is terrible. Its only redeeming characteristic is that it is better than first-past-the-post.

I prefer a top two runoff with approval voting. They should all face off in a single non-partisan primary and the two with the highest approval should face off in the general election. It is simple and would provide a much need change in how political parties should work. Parties should have no say or special privilege in getting candidates on the ballot. Every candidate should have to get the same number of signatures to be put on the primary ballot regardless of party affiliation.

The function of political parties should be changed to act as an advocacy group and a unified voting bloc to support all the candidates that share their view. Allow candidates to be supported by multiple parties and parties to support multiple candidates. In order to eke out a victory ,candidates would need to focus on the most important issues and win the support from as many popular parties as possible. This would provide a powerful counterbalance to money in politics. This would be the best method to give third parties any leverage.

 

[Jayjay]

I'm skeptical of approval voting because it gives room for tactical voting. If you have one favourite, you can increase his chances for winning by putting non-approval for everyone else, including those who you would consider acceptable. So the best voting strategy with approval voting is still to just give one vote and it degenerates to FPTP... if it wasn't for people who have no opinions or don't understand the tactical aspect of it.

For single seat districts that make up a larger assembly or legislature, I would actually like to see random ballot: if a candidate gets 50% of the vote, he would have 50% chance of winning. If he gets 1% of votes he has 1% chance of winning. That way every vote would count.

From any realistic options run-off is usually good enough as it allows 3-4 serious candidates rather than just 2, and IRV or condercet could be used to ake it so you only have to vote once.

 

[Blahface]

I'm skeptical of approval voting because it gives room for tactical voting. If you have one favourite, you can increase his chances for winning by putting non-approval for everyone else, including those who you would consider acceptable. So the best voting strategy with approval voting is still to just give one vote and it degenerates to FPTP... if it wasn't for people who have no opinions or don't understand the tactical aspect of it.

You can always vote for your favorite without worry though. If you are worried about the candidate you hate most winning, you can compromise and vote for other acceptable candidates as well. Under IRV, you can actually hurt your favorite by voting for him.

If you have a runoff with approval, you have a cushion and you can vote more honestly. In the first round, you can vote on issues. On the second round, you can vote for the best candidate to implement those issues.

 

[Jimmy Higgins]

The thing I don't like about rank choice voting(especially IRV) is that too many people just don't seem to understand it. Even if you can come up with a perfect Condorcet method in which there is no strategic benefit to burying your favorite candidate's top rival, you have to get the average voter to understand that. It might be tough to get the average voter to even understand what a Condorcet method is in the first place. Also, IRV is terrible. Its only redeeming characteristic is that it is better than first-past-the-post.

I prefer a top two runoff with approval voting. They should all face off in a single non-partisan primary and the two with the highest approval should face off in the general election. It is simple and would provide a much need change in how political parties should work. Parties should have no say or special privilege in getting candidates on the ballot. Every candidate should have to get the same number of signatures to be put on the primary ballot regardless of party affiliation.

The function of political parties should be changed to act as an advocacy group and a unified voting bloc to support all the candidates that share their view. Allow candidates to be supported by multiple parties and parties to support multiple candidates. In order to eke out a victory ,candidates would need to focus on the most important issues and win the support from as many popular parties as possible. This would provide a powerful counterbalance to money in politics. This would be the best method to give third parties any leverage.

What stops a Republican leaning area to Vote their top guy and then the Green party guy in? Or visa versa, the top Democrat and then the Libertarian? This is America... if we can fuck up a good way of doing something... we will!

 

[Blahface]

The thing I don't like about rank choice voting(especially IRV) is that too many people just don't seem to understand it. Even if you can come up with a perfect Condorcet method in which there is no strategic benefit to burying your favorite candidate's top rival, you have to get the average voter to understand that. It might be tough to get the average voter to even understand what a Condorcet method is in the first place. Also, IRV is terrible. Its only redeeming characteristic is that it is better than first-past-the-post.

I prefer a top two runoff with approval voting. They should all face off in a single non-partisan primary and the two with the highest approval should face off in the general election. It is simple and would provide a much need change in how political parties should work. Parties should have no say or special privilege in getting candidates on the ballot. Every candidate should have to get the same number of signatures to be put on the primary ballot regardless of party affiliation.

The function of political parties should be changed to act as an advocacy group and a unified voting bloc to support all the candidates that share their view. Allow candidates to be supported by multiple parties and parties to support multiple candidates. In order to eke out a victory ,candidates would need to focus on the most important issues and win the support from as many popular parties as possible. This would provide a powerful counterbalance to money in politics. This would be the best method to give third parties any leverage.

What stops a Republican leaning area to Vote their top guy and then the Green party guy in? Or visa versa, the top Democrat and then the Libertarian? This is America... if we can fuck up a good way of doing something... we will!

Do you mean trying to vote for a weaker candidate so that your favorite faces off against a weaker opponent in the general? That would be a pretty bad strategy. There is always the chance that you can push your favorite candidate out completely.

 

[dx713]

Under IRV, you can actually hurt your favorite by voting for him.

How?
Can you clarify?

 

[Blahface]

Under IRV, you can actually hurt your favorite by voting for him.

How?
Can you clarify?

Lets say three candidates A,B, and C are running for office using IRV. The voters prefer the candidates in the following order:

24: A C B
21: B A C
20: C B A

C is eliminated and B wins. If two of the A voters had made A their second choice and C their first choice, B would have been eliminated and A would have won in-spite of receiving a smaller number of first choice votes .

 

[dx713]

How?
Can you clarify?

Lets say three candidates A,B, and C are running for office using IRV. The voters prefer the candidates in the following order:

24: A C B
21: B A C
20: C B A

C is eliminated and B wins. If two of the A voters had made A their second choice and C their first choice, B would have been eliminated and A would have won in-spite of receiving a smaller number of first choice votes .

Thanks, I hadn't thought of that.
But I note that this is a case where there is no Condorcet candidate. A majority prefers B over A, but a majority prefers C over B, and a majority prefers A over C.
Any counting method you choose here will not give you an ideal outcome, because there is simply no ideal outcome to be had.
So I still think IRV is the best representativity / simplicity compromise. It's already on the complicated side for most people, but it can be explained and educated (e.g., students representative and class representative could be elected with that, with manual counting, that would demistify the system for the new generation), and the less complicated systems (two-turns, approval), are too open to political manipulation for me.

 

[lpetrich]

I checked out Blahface's example with Voting Calculator. I entered

24:A>C>B
21:B>A>C
20:C>B>A

and calculated everything.

In that page, basic Condorcet is apparently changing the smallest defeats to zero until one gets a Condorcet winner. Beatpath is the Schulze algorithm and ranked pairs the Tideman algorithm. All three agreed on A being the winner.

IRV, however, has B as the winner.

Math Alive: Voting & Social Choice — Lab 1 has a nice example of how different methods can give different results: FPTP, top-two, IRV, Borda, and Condorcet. Its examples all have Condorcet winners for simplicity.

Borda: lowest-ranked candidate gets a 1, the next one gets a 2, ..., and these numbers get added up for all the ballots.

Condorcet: virtual round robin. One turns the preferences into a Condorcet matrix, a table of how often which candidate beats which other candidate. A Condorcet winner is one who beats every other one in these pairwise contests. If there is no such winner, then one can find the winner from the Condorcet matrix using any of some rather complicated algorithms.

.

Schulze beatpath. Create a matrix of beatpath strengths for each pair of candidates. Each pair has several paths between them that can go through the other candidates. The strength of each path is the size of the weakest victory along it, and the overall strength is the maximum of these strengths. There is algorithm for calculating the beatpath matrix that is based on the Floyd-Warshall algorithm that runs in O(n³) for n candidates. Once one has it, one can find the beatpath winner by doing what one does to find the Condorcet winner for the original matrix.

Tideman ranked pairs. Turn the Condorcet matrix into a list that is sorted from maximum to minimum victory strength. Select each pair and its strength starting from the maximum one and going downward. If the resulting set of rankings has a cycle, then remove that added pair and continue to the next one. One can look for cycles with this method. Look for every listed candidate that beats without being beaten. Look for every one that gets beaten without beating any one. Remove all rankings that contain those candidates. Repeat until one cannot proceed any further. If none are left, then there are no cycles.

Kemeny-Young. Find all permutations of candidates. For each one, add up how much each one beats each later one. The permutation with the highest total wins. This is O(n!), and for a large number of candidates, one may have to do something like simulated annealing.

Minmax. Find the maximum defeat for each candidate, and the winner is the candidate with the smallest one. One can use the Condorcet matrix itself for the defeats (pairwise opposition), the difference between the matrix value and the opposite-direction value (margins), or the matrix value only if it is greater than the opposite-direction value, otherwise zero (winning votes).

Nanson's method. Do the Borda count. Then remove the candidates with below-average scores and repeat until one candidate is left.

Baldwin's method. Do the Borda count. Then remove the lowest-scoring candidate and repeat until one candidate is left.

 

[lpetrich]

Top-two runoff is an improvement over first-past-the-post, but it can have screwy outcomes.

Some US states use it in the form of a blanket primary or jungle primary: Louisiana, California, and Washington. However, it's been defeated in Arizona and Oregon. Several nations use it to elect their presidents, notably France.

In those states, it has sometimes produced two Democrats or two Republicans advancing to the main election. For instance, in the recent election, Sandra Fluke finished second in the primary for the California State Senate in Santa Monica, but she lost to a fellow Democrat in the main election. In France in 2002, far-right candidate Jean-Marie Le Pen beat Socialist Lionel Jospin and advanced to the runoff, along with conservative Jacques Chirac of the Rally for the Republic. However, Le Pen lost to Chirac in the runoff, getting only a few more votes than in the first round.

 

[dx713]

Top-two runoff is an improvement over first-past-the-post, but it can have screwy outcomes.

Some US states use it in the form of a blanket primary or jungle primary: Louisiana, California, and Washington. However, it's been defeated in Arizona and Oregon. Several nations use it to elect their presidents, notably France.

In those states, it has sometimes produced two Democrats or two Republicans advancing to the main election. For instance, in the recent election, Sandra Fluke finished second in the primary for the California State Senate in Santa Monica, but she lost to a fellow Democrat in the main election. In France in 2002, far-right candidate Jean-Marie Le Pen beat Socialist Lionel Jospin and advanced to the runoff, along with conservative Jacques Chirac of the Rally for the Republic. However, Le Pen lost to Chirac in the runoff, getting only a few more votes than in the first round.

As a French, it was the first time of my life I voted for the right-wing candidate!

The problem with two-turns elections is that it gives way to a lot of political manoeuvering. In the example you cite, Lionel Jospin wasn't absent from the second round because he had less approval than Le Pen. But because everybody wanted to express their ideas instead of voting "useful" (including me, to be honest), so his support was split between him, the radical left, the greens, the ex-communists, etc.
(and it's important to express your ideas, because it will often lead to one of the second round candidates to make promises to secure your vote, if they're supported strongly enough)

The possibility of this kind of scenario is why you end with plenty of backroom discussions going from the reasonnably democratic "don't run and I'll include this line from your program in mine" (although it is democracy through opinion polls instead of democracy through votes, so already iffy), to the "don't run and I'll have you a nice cushy place in my administration if I win" arrangements.

And why I would advice any USian with a voice in reforming their voting system to NOT look to France but at the very least to approval voting.

 

[dx713]

And I see in your lab link that you can have examples where the IRV winner is not the Condorcet winner.
So, maybe IRV isn't a system I can support?
On the other hand, I don't fancy explaining Schultz to anyone without serious mathematics education.
(not that the process is difficult to explain, there are nice graphical methods, but answering the "why do you do it that way" and "are you sure" questions, when you see how the UK opposition to IRV already managed to misrepresent the system...)

Maybe a Black (Condorcet+Borda) method? Or a Condorcet+IRV?
Well that doesn't change my vote in the poll, it's a preferential voting system anyway.

 

[Jayjay]

Top-two runoff is an improvement over first-past-the-post, but it can have screwy outcomes.

Some US states use it in the form of a blanket primary or jungle primary: Louisiana, California, and Washington. However, it's been defeated in Arizona and Oregon. Several nations use it to elect their presidents, notably France.

In those states, it has sometimes produced two Democrats or two Republicans advancing to the main election. For instance, in the recent election, Sandra Fluke finished second in the primary for the California State Senate in Santa Monica, but she lost to a fellow Democrat in the main election. In France in 2002, far-right candidate Jean-Marie Le Pen beat Socialist Lionel Jospin and advanced to the runoff, along with conservative Jacques Chirac of the Rally for the Republic. However, Le Pen lost to Chirac in the runoff, getting only a few more votes than in the first round.

As a French, it was the first time of my life I voted for the right-wing candidate!

The problem with two-turns elections is that it gives way to a lot of political manoeuvering. In the example you cite, Lionel Jospin wasn't absent from the second round because he had less approval than Le Pen. But because everybody wanted to express their ideas instead of voting "useful" (including me, to be honest), so his support was split between him, the radical left, the greens, the ex-communists, etc.
(and it's important to express your ideas, because it will often lead to one of the second round candidates to make promises to secure your vote, if they're supported strongly enough)

The possibility of this kind of scenario is why you end with plenty of backroom discussions going from the reasonnably democratic "don't run and I'll include this line from your program in mine" (although it is democracy through opinion polls instead of democracy through votes, so already iffy), to the "don't run and I'll have you a nice cushy place in my administration if I win" arrangements.

And why I would advice any USian with a voice in reforming their voting system to NOT look to France but at the very least to approval voting.

It should be obvious that adding a run-off stage just moves the spoiler effect one rung lower. So if in FPTP you get two serious candidates, with run-off vote you get at least 3. Three is better than two. Maybe approval voting for the first round would be even better.

 

[Blahface]

As a French, it was the first time of my life I voted for the right-wing candidate!

The problem with two-turns elections is that it gives way to a lot of political manoeuvering. In the example you cite, Lionel Jospin wasn't absent from the second round because he had less approval than Le Pen. But because everybody wanted to express their ideas instead of voting "useful" (including me, to be honest), so his support was split between him, the radical left, the greens, the ex-communists, etc.
(and it's important to express your ideas, because it will often lead to one of the second round candidates to make promises to secure your vote, if they're supported strongly enough)

The possibility of this kind of scenario is why you end with plenty of backroom discussions going from the reasonnably democratic "don't run and I'll include this line from your program in mine" (although it is democracy through opinion polls instead of democracy through votes, so already iffy), to the "don't run and I'll have you a nice cushy place in my administration if I win" arrangements.

And why I would advice any USian with a voice in reforming their voting system to NOT look to France but at the very least to approval voting.

It should be obvious that adding a run-off stage just moves the spoiler effect one rung lower. So if in FPTP you get two serious candidates, with run-off vote you get at least 3. Three is better than two. Maybe approval voting for the first round would be even better.

Yeah, I wanted to pull my hair out when Cenk Uygur was moderating a debate for Henry Waxman's old seat and he said, “We have a top two system, so you can vote for whoever you want and not worry about it.” After the election, he got a little wise about and it and said, “Oh, well it turns out that the progressives can still split the vote.” That is the first time I have heard him talk about voting systems in any form and since then he has never went on to talk about it again.

A top two system needs approval voting. The reason I like it so much is because it wouldn't just get better candidates, but it would change the nature of political parties.